[Webkit-unassigned] [Bug 48308] Too precise serialization from floating point number to string for "number" input elements

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=48308





--- Comment #24 from Dai Mikurube <dmikurube at google.com>  2010-11-01 05:25:47 PST ---
(In reply to comment #23)
Hi Gavin,

Ok, I agree with your final suggestion. Rounding in applyStep() is ideally the best and your rounding method looks great.

But I think it requires deep modifications in the class WTFString and other fundamental classes. (E.g. adding another overloaded WTFString::toDouble(bool *, int decimalPlaces), the same for some StringImpls, parseToDoubleForNumberType, InputType::parseToDouble, *InputType::parseToDouble, and so on...)

Is it acceptable? I'd like to hear some opinions from others. It was the reason that I was hesitating acculate rounding in applyStep(). It essentially requires information of the given string.


Memo:

I thought maximum/minimum limitations for confirmation, but no problem.
For single-precision numbers, DP is at most 38. The maximum number in single precision is 10^38. 10^(38*2) is ultimately required temporally. Double-precision can contain at most 10^308. (* But, of course, rounding errors are large near the boundaries.)

But we have to limit DP in parsing unreasonably long numbers given as a string. (E.g. <input step="0.0000...(snip)...01">)

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.